Home / Ideas Letter

Pension Risk – Modelling and Quantifying Risk

This article discusses model risk and quantification of risk from a pension plan perspective.

Ideas Letter · Vol. 34 · Oct 2012

Background

The maxim “all models are wrong, but some are useful” is usually quoted when discussing model risk. We do not subscribe to this notion since we believe that quantification of risk requires development of a model which correctly interpreted is the “quantitative abstraction of reality”. Models were never supposed to be 100% reflective of reality! Ideally, any model should be simple, straightforward, robust, easily interpretable and explainable– unnecessary complexity in a model should be avoided if such complexity is not clear and not easily understood (note: the 2007-2008 credit crisis was a result of not understanding the limitations of models being used). In Einstein’s words “Everything should be made as simple as possible, but not simpler”.

Modelling and Quantifying Risk

Pension modelling, that is, developing a model to analyse the dynamics of a pension plan must start with deciding on the risks that need to be modelled and the desired output that is required to make specific decisions, for example, on key funding or accounting measures, contribution rates, or solvency measures. Pension modelling should not be limited to Asset Liability Study. It is important that a pension model is developed to avoid the following risks: selection risk( having a correct underlying pension model framework), construction risk(introducing structured construction processes with peer review), parameter risk (understanding inputs and impact on output) and interpretation risk(understanding output and impact on decision making) – these risks are collectively termed “model risks”. For a pension plan, at a minimum, the following inputs must be included in a pension model: Interest (discount) rates, salary growth, inflation, equity premium, demographic (early retirement, termination, disability, etc) assumptions and mortality (life expectancy). If the pension plan is managed using an ERM framework, then output should be reviewed incorporating the following considerations: sponsor covenant risk ( financial strength of the plan sponsor to meet its obligation from a downside perspective), economic benefit ( upside potential for stakeholders and who will benefit – plan sponsor or beneficiaries) and business sensitivities(correlation of an economic exposure that could be a natural hedge for the business, or compound a risk exposure resident elsewhere in the business).

Ideas in Action

The best response to modelling and quantifying risk is to reject the notion of avoiding models. Model risk can be minimized by introducing the sound practice of having a separate team developing the model and another team interpreting the model- discourse and tension between the two teams will result in a structured and robust model with a better understanding of the limitations of the model. Supplemental Reading: Lalani Ideas Letter April 2012, Volume 28.

Written for professional readers and reflects my views at the date shown. It is not actuarial, investment or legal advice.